Jac VerSteeg: Poll accuracy might be worse than you thought

1

Way back in 1984, the late, great Chicago newspaperman Mike Royko wrote a column in which he urged readers to lie to pollsters.

“Be polite. Talk to them,” Royko coached. “But lie. Don’t give them an honest answer.” Royko wrote that pollsters would collect the results and “feed them into a computer, which will chew on them, digest them and finally burp a sheet of paper” allowing the network’s “high priest of politics” to project the winner.

“But if enough of you lie,” Royko encouraged, “the entire nation will be treated to one of the finest evenings of television viewing since the tube was unleashed.”

Here we are more than 30 years later and it is becoming increasingly clear that people don’t need to lie to pollsters to screw up the results. Because of changes in the way human beings communicate – and I’m talking here primarily about smart phones – it might be impossible to conduct a valid poll. In which case the pollsters, some completely unwittingly, are lying to us.

Certainly there have been a lot of incorrect predictions recently. Results in Kentucky, where pollsters overwhelmingly and erroneously predicted a Democrat would be the next governor, are just the latest high-profile example.

Bob Sparks explained in his recent column for Context Florida that poor methodology can be partly to blame.  And Sparks cited an analysis by Steve Vancore, also for Context Florida, that bolstered the point.

Both of those columns are must reading for anyone trying to figure out how two polls – each supposedly the result of scientific research – can come to different conclusions or get election results so wrong.

And polls do often disagree. As reported in the Nov. 8 South Florida Sun Sentinel, the NBC News/Wall Street Journal survey of Republican voters found Ben Carson on top with 29 percent to Donald Trump’s 23 percent, while the Quinnipiac University poll put Donald Trump on top with 24 percent and Ben Carson second with 23 percent.

The discrepancy might be a simple matter of timing. Those two polls overlapped slightly on the calendar but were not conducted on identical days. They also differed in sample size, which slightly affects the margin of error.

However, voices are emerging from the world of polling that sound a more serious alarm.

In a story in September, Bloomberg News reported that “doubts are intensifying (about the accuracy of polling) after a series of high-profile misfires around the world in the past year, notably in Greece, Israel and Britain.”

The key problem – which some also blame for the flawed polling that convinced Mitt Romney he would defeat Barack Obama in 2012 – is a decline in participation traced to the rise of cell phones.

People increasingly do not have land lines and do not answer their cell phones unless they recognize the caller. Even then, they are as likely to text as call. Or they could be at the mall or the grocery store and decline to talk.

Further, mobile phones are, well, mobile, which could have an effect on local or regional polling. In my own family, three people whose phone numbers have Florida area codes no longer reside in the Sunshine State.

The Bloomberg piece quotes Kirby Goidel, editor of “Political Polling in the Digital Age”: “There isn’t a pollster out there who thinks about this seriously who isn’t a little bit uneasy.”

Bloomberg is far from the only source reporting the concern. U.S. News & World Report,  in a story from September,  quotes Geoffrey Skelley of the University of Virginia’s Center for Politics: “The science of public surveying is in something of a crisis right now.”

All the uncertainty leads to the inevitable questions. Are Trump’s and Carson’s numbers really that good? Are Jeb Bush’s numbers really that bad? There is no definitive answer since political polls, by their very nature, cannot be tested until there is an actual election.

Even when we get to the caucuses and primaries – starting in early February of next year – we never will know whether the polls from this fall were correct.

Because polling is “in crisis,” 2016’s elections will be the most important test for polling since public opinion surveys began. Not only will voters decide which candidates will hold office, they will learn which polling firms to trust – if any.

In the meantime, when considering polls, remember that they might reflect built-in lies that make Royko’s suggestion seem quaint.

Jac Wilder VerSteeg is a columnist for The South Florida Sun Sentinel, former deputy editorial page editor for The Palm Beach Post and former editor of Context Florida. 

Share.

1 Comment

Show Buttons
Hide Buttons
%d bloggers like this: